Variability in Second Language Article Production: A Comparison of L1 Thai and L1 French Learners of L2 English

This thesis explores variability in second language (L2) production of English
articles by speakers whose first language (Ll) is Thai (-articles), and
compares it with proficiency-matched learners whose Ll is French (+articles).
The thesis addresses a current debate on whether variability in production of
second language functional morphology stems from representational deficits
or from processing problems in production.
The investigation of L2 article production was focused on tightly
defined pairs of contexts for which different theoretical positions would
predict different learner behaviours. Experiments were designed which
measured the level of article omissions (a) in adjectivally premodified noun
phrases (Art + Adj + N) vs. non-modified phrases (Art + N), (b) with first vs.
second mention definite referents, and (c) with more vs. less attended (less
salient) referents. A further study explored article substitution errors, in
particular the alleged overuse of the indefinite article in [+definite; -specific]
contexts and the definite article in [-definite; +specific] contexts.
Results suggest that Ll Thai learners of L2 English, but not Ll
French learners of L2 English, omit articles more (a) in adjectivally
premodified than in non-modified contexts, (b) with second than with first
mention definite referents, and (c) with more attended than with less attended
referents. It is argued that these results point against the view that variability
in production of L2 morphology stems from processing problems in
production only (i.e. the view that assumes that L2 syntax must be target-like),
and that they support the view that the variability stems from representational
problems, with further knock-on effects on processing. In particular, the
results are interpreted as consistent with the combined predictions of the
Syntactic Misanalysis Hypothesis (Trenkic 2007) and the Information Load
Hypothesis (Almor 1999). The results of the study investigating article substitutions show that
Ll Thai learners of L2 English, but not Ll French learners of L2 English,
oversupplied the definite article in fill-in-the-gap tasks in [-definite; +specific]
contexts, but only when the speaker explicitly claimed personal familiarity
with the referent - not when the speaker explicitly denied persona' familiarity
with the referent. Similarly, they also oversupplied the indefinite article in
[+definite] contexts whenever the speaker denied familiarity with the referent,
irrespective of whether the context was [+specific] or [-specific]. This
suggests that LI Thai learners of English accept familiarity with identifying
attributes of a referent as a possible criterion for the use of the definite article,
and non-familiarity as a criterion for the use of the indefinite article. The
results are interpreted as arguing against the suggestion that L2 English article
choices are UG-regulated (cf. lonin, Ko and Wexler 2004). The results are
shown to be consistent with the predictions of the Syntactic Misanalysis
Hypothesis instead.
The results of empirical investigations conducted in this thesis
contribute to the debate on causes ofvariability in production ofL2 functional
morphology. The observed patterns of L2 English article omissions and
substitutions seem more consistent with the view that variable production
stems from non-target-like syntax.